Research Software Engineer · IBM Research
Maia Iyer
Building secure-by-default, identity-first systems for zero trust, agentic security, and multicloud.
About
I'm an Advisory Software Engineer at IBM Research with over four years of industry research experience with special interest in distributed systems, security, and agentic systems primitives.
I'm a maintainer of the Kagenti (rossoctl) project towards platform primitives that establish a zero-trust architecture. I'm also a maintainer of the open-source Tornjak project (SPIFFE community) where I help drive the adoption of SPIRE as an open-source identity implementation for secure cloud-native ecosystems.
I'm deeply passionate about technical education and community engagement. Whether speaking at conferences, designing hands-on workshops, or writing architectural deep-dives, I view communication as a core engineering responsibility - empowering the greater ecosystem with actionable ideas to build trustworthy systems.
Highlighted works
-
Talk
When an Agent Acts on Your Behalf, Who Holds the Keys?
KubeCon EU 2026 — binding agent identity to delegated user identity with SPIRE, Keycloak, and an MCP Gateway.
-
How OAuth 2.0/2.1 underpins MCP authorization — least-privilege, short-lived tokens instead of static agent credentials.
medium.com Security in and around MCP: Part 1 - OAuth in MCP Least-privilege OAuth for MCP — short-lived tokens instead of static agent credentials. -
Podcast
Your AI Agent Has Too Much Access | Fixing Agent Identity
Kubernetes Bytes — user vs. workload identity, OAuth, SPIFFE/SPIRE, Keycloak, and identity-first patterns for agents.
Conference talks
Upcoming
-
KubeCon NA · 2026Your CI's Mistaken Identity [event]
-
KubeCon NA Open Source SecurityCon · 2026License to Generate: Securing AI Agents with Keycloak, SPIRE, and Envoy [event]
-
KubeCon NA Platform Engineering Day · 2026Don’t Lose Your Agent’s Mind: Platform Patterns for Stateful Agents [event]
Previous
-
Kubernetes Bytes · 2026Your AI Agent Has Too Much Access | Fixing Agent Identity [recording]
-
Agentic AI Summit · 2026From Sandbox to Swarm: The Missing Platform Blueprint for Production-Ready Agents [poster]
-
KubeCon EU · 2026When an Agent Acts on Your Behalf, Who Holds the Keys? [recording]
-
KubeCon EU KeycloakCon · 2026Advancing Machine-to-Machine Workflows with the SPIFFE Identity Provider [recording]
-
KubeCon NA · 2025Build-a-Bot Workshop: Enabling Trusted Agents With SPIRE + MCP [recording]
-
KubeCon NA Open Source SecurityCon · 2025Who Let the Agents Out? Securing AI Workflows the Right Way [recording]
-
Workload Identity Day London · 2025Zero-trust Based Workload Identity Solutions for Agentic Platforms [recording]
-
Workload Identity Day Salt Lake City · 2024Federating Trust in the Cloud Multiverse with SPIRE and Tornjak
-
CloudNativeSecurityCon NA · 2024Demystifying and Enabling Workload Identity Across the Cloud Native Ecosystem [recording]
Technical Publications
Peer-reviewed papers
- arXiv · 2024
- ICML KLR Workshop Poster Paper · 2024
- IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · 2018
Medium
- 2025
- 2025
- 2025
- 2022
- 2022
Experience
-
Master of Science in Computer Science Carnegie Mellon University · School of Computer Science
Sep 2021 — May 2022 -
Hybrid Cloud Research Intern IBM T.J. Watson Research Center
May 2021 — Aug 2021 -
Bachelor of Science in Computer Science Carnegie Mellon University · School of Computer Science
Sep 2017 — May 2021Minor in Logic and Computation
-
Hybrid Cloud Research Intern IBM T.J. Watson Research Center
Jun 2020 — Aug 2020 -
NSF REU Research Intern University of Connecticut
May 2019 — Aug 2019 -
NSF REU Research Intern IUPUI
May 2018 — Aug 2018
